In this study, we investigated the correlation between multifidus muscle atrophy, facet joints, and disc degeneration in LDH patients. The result of this study supported our hypothesis that in patients with L4-L5 LDH, multifidus muscle grading was higher at the L5-S1 level than at the L3-L4 level. Positive correlation was identified between LDD and LMA at the L3-L4 level, but no correlation was found at the L5-S1 level.
We found that compared with the control group, herniated patients had more severe LMA at the three levels. There were two possible explanations. First, it could be because the selected patients in the control group had no obvious disc degeneration symptoms. Second, multifidus muscle disuse caused by specific back pain and denervation caused by nerve root compression were the main reasons responsible for this phenomenon.
MRI results from this study demonstrated that there was more severe LMA at the L5-S1 level than at the L3-L4 level in L4-L5 LDH patients. Anatomically [15, 16], multifidus fascicles arise from the spinous process and adjacent lamina of each lumbar vertebra, descend caudolaterally, traverse L3-L4 vertebral levels, and is innervated by the nerve root of the same level. Similar to the L5 level, multifidus muscle arising from the L5 spinous process is innervated by the L5 nerve root. In patients with L4-L5 LDH, if the L5 root was affected, muscle denervation would be expected only at the L5 level. The anatomical features result in more muscle volume at the inferior segment than the superior segment.
Two mechanisms for muscle atrophy have been proposed: disuse and denervation. In animal studies [17, 18], it is generally agreed that denervation leads to decrease in Type II fiber size. In contrast, opinion differs concerning changes in the Type I fiber size after denervation. However, Yoshihara et al.  demonstrated that denervation caused by nerve root compression may lead to atrophy of both fiber types (type I and type II) with structural changes in the human lumbar multifidus muscle.
Despite these controversies, we proposed two mechanisms that are possibly responsible for multifidus atrophy. In LDH patients, disuse/immobilization of the back muscles is common. Such changes may explain the generalized effect (atrophy) at five levels. In addition, symptoms including paraspinal denervation (short-angled fibers) and re-innervation (grouped fibers, polyphasic action potentials) are common in disc herniation and nerve root compression. The CSA of multifidus muscle innervated by medial branch of the dorsal ramus of the L5 nerve root are reduced when L5 nerve root is compressed by herniated mass.
Numerous studies have reported rapid muscle atrophy in response to inflammation [20, 21] and muscle or joint injury . However, the mechanisms for such changes are poorly understood. Paraspinal muscle plays a more important role in protecting the L3-L4 segment than others. The multifidus muscle maintains the lumbar lordosis by acting like a bowstring which helps to transmit some of the axial compression force on the disc to the anterior longitudinal ligament by switching compression loading to stretch loading; additionally, it protects the discs by preventing unexpected movements like torsion and flexion . This phenomenon not only requires sufficient strength muscle, but also completes the nucleus pulposus and annulus fibrosus structure. For the L5-S1 segment, however, the intervertebral disc below the iliac crest, the ilium could restrict segment movement, thus reduce the pressure and shear forces on the intervertebral disc.
We found no correlation between L5-S1 level disc degeneration and multifidus muscle atrophy in our study. At the L3-L4 level, as there was no denervation phenomenon, the positive correlation demonstrated that LMA could be the cause of disc degeneration. At the L5-S1 level, however, the differences of LMA between the two groups suggested that LMA could also be the consequence of L4-L5 disc herniation.
This study analyzed the lumbar degenerative changes between herniated and adjacent segments in L4-L5 herniated disc patients using MRI. The multifidus muscle plays a critical role in controlling and stabilizing the lumbar spine in multiple planes of action, including axial rotation as well as posterior rotation (extension) on the sagittal plane .
Lower back pain can recollapse more often in case of weakened paraspinal muscles resulting in instability; the latter can be improved by rehabilitation exercise programs [24, 25]. For disc-herniated patients, microdiscectomy results in minimal atrophy and fatty infiltration to the multifidus muscle [26–29]. And lumbar extension muscle-strengthening program can decrease paraspinal muscle atrophy after lumbar disc herniation surgery, and prevent adjacent level degeneration and lumbar spondylolisthesis, especially for patients with severe LMA caused by lower lumbar range of motion (ROM) after posterior lumbar interbody fusion.
Choi et al.  demonstrated that pain, duration of recuperation, and back-muscle strength can be affected favorably by lumbar extension muscle-strengthening programs after lumbar disc herniation surgery. Thus, it is recommended that MRI should be performed to obtain a description of LMA in disc-herniated patients, and clinicians should pay attention to LMA when instituting lumbar extension muscle strengthening programs.
There are several limitations with this study. First, we analyzed only patients with mono-segment disc herniation at L4-L5; excluding L3-L4/L5-S1 disc herniation and multi-segment degeneration due to the relatively small sample size. Second, we were not able to perform a comparative analysis to investigate the difference between the herniated cohort and control groups. Further study with a larger number of cases should be performed to obtain an age- and sex-matched control result and thus confirm the results of the present study.